Spare Parts[1]

8/2/2019 Spare Parts[1]

1/21

nternational Journal of Physical Distribution & Logistics Managementmerald Article: Spare parts optimization process and results: OPUS10ases in the Norwegian Defence

ernt E. Tysseland

rticle information:

o cite this document: Bernt E. Tysseland, (2009),"Spare parts optimization process and results: OPUS10 cases in the Norwegian

efence", International Journal of Physical Distribution & Logistics Management, Vol. 39 Iss: 1 pp. 8 - 27

rmanent link to this document:

p://dx.doi.org/10.1108/09600030910929165

ownloaded on: 17-04-2012

eferences: This document contains references to 17 other documents

o copy this document: [email protected]

ccess to this document was granted through an Emerald subscription provided by SYMBIOSIS INTERNATIONAL UNIVERSITY PU

or Authors:

you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service.

formation about how to choose which publication to write for and submission guidelines are available for all. Additional help

r authors is available for Emerald subscribers. Please visit www.emeraldinsight.com/authors for more information.

bout Emerald www.emeraldinsight.com

ith over forty years' experience, Emerald Group Publishing is a leading independent publisher of global research with impact in


2/21

Spare parts optimization processand results

OPUS10 cases in the Norwegian Defence

Bernt E. TysselandRoyal Norwegian Naval Academy and Molde University College, Bergen, Norway

Abstract

Purpose The paper has two main aims: to focus on how the spare parts optimization process wasconducted in the Norwegian Defence procurement projects that had used the system approach basedon OPUS10, and whether coordination issues affected the process and results; and to analyse empiricaldata in order to evaluate whether the theoretical claim of the system approach used through OPUS10,being better than other methods in terms of availability and spare parts investment cost holds up inreality.

Design/methodology/approach Both qualitative and quantitative methods were used in order toanswer the different questions of the study.

Findings Very few Norwegian Defence projects have used the system approach through OPUS10.Empirical data however comply with the theoretical claims of potential large savings in spare partsinvestment cost and/or improvement in operational availability. Several organizational factors canexplain the lack of use of OPUS10. The most important being lack of resources, lack of a centralizedconcept and a somewhat low-project leader attitude towards the approach.

Research limitations/implications The study of Norwegian Defence cases makes

generalizations of findings not applicable. The research model could however easily be transferredand utilized in the study of other organizations spare parts optimization processes.

Practical implications The Norwegian Defence should alter their concept for project governanceand management in order to gain the full potential of the system approach used through OPUS10.

Originality/value Few research papers have evaluated the promising theoretical findings ofsystem-based optimization based on empirical operational data. Even fewer, if any, studies have useda combination of factors from organization theory, economic organization theory and operationmanagement theory in order to explain findings based on predefined hypotheses. This research shouldhave value for both practitioners and researchers within the field spare parts optimization in generaland systems management in particular.

Keywords Spare parts, Norway, Defence sector, Optimization techniques, Project planning

Paper type Research paper

1. Introduction and research questions

Approximately, 33 per cent of the annual defence budget in Norway[1] is connected toinvestments (PRINSIX, 2008). In an attempt to secure cost effective use of this largeamount of money the Norwegian Defence has developed a concept for projectgovernance and management called PRINSIX (2008). PRINSIX was given authority asthe official project management system in 1996. PRINSIX is now based upon theinternational acclaimed Project Management Body of Knowledge (see www.pmi.org) inaddition to the unique Norwegian military experience. PRINSIX consist of four mainareas, with maybe the most important area being the methodological approachconsisting of a decision model, a governance model, knowledge areas and documenttemplates. One of the knowledge areas is integrated logistic support (ILS). ILS was first

The current issue and full text archive of this journal is available atwww.emeraldinsight.com/0960-0035.htm

IJPDLM39,1

8

nternational Journal of PhysicalDistribution & Logistics ManagementVol. 39 No. 1, 2009

p. 8-27q Emerald Group Publishing Limited

960-0035

DOI 10.1108/09600030910929165


3/21

developed by the US Department of Defence because their military material projectshad been completed later than planned and at a higher cost than budgeted (Kumar et al.,2000). ILS is a tool for making sure that the cost of operating, servicing and retiringequipment can be kept at a minimum in the same time as equipment performancerequirements are met (Blanchard, 2004). In order to achieve this goal of bigger bangfor the buck, the ILS method is divided into ten elements[2] that separate the logisticchain into manageable chunks (Farmer et al., 2003). Spare parts provisioning andmanagement is one of the most challenging problems in the whole ILS process (Kumaret al., 2000). Earlier Norwegian Defence directives have stated that approximately 5 percent of total system investments should be used for spare parts covering

approximately two years of operational needs (PS2000, 1995). In a research projectfrom 1995, it was shown that Norwegian Defence procurement projects have madeinvestments in spare parts that are never used. The reason for this is, among otherthings, that they have accepted system suppliers spare parts suggestions withoutevaluating the suggestion based on the so-called multi-echelon, multi-item,multi-indenture method (PS2000, 1995). In the same research, the NorwegianDefence was recommended to start using the multi-echelon, multi-item, multi-indentureoptimization software tool called OPUS10 (see www.systecon.se) (PS2000, 1995).More than ten years have gone by since the Norwegian Defence introduced PRINSIX asthe official project management system and since OPUS10 was chosen as the spareparts optimization tool. In the same time period there has been a growing attention formanagement and organizational-related issues both in academia and industry.Nevertheless, the area of inventory management still seems to lack a clear linkagebetween planning and control issues on one hand and organizational issues on the

other hand (de Vries, 2005). It is believed that in the spare parts process of the differentprocurement projects, organizational issues such as coordination problems both withinthe organization and towards the supplier might affect both the process and the finalresult. Based upon the above, the first research question becomes:

RQ1. How has the spare parts optimization process been conducted in theNorwegian Defence procurement projects that has used OPUS10, and hascoordination issues affected the process and results?

Secondly:

RQ2. Does empirical data show that the multi-echelon, multi-item, multi-indenturemethod solved through OPUS10, improves system availability and/or sparesparts investment cost compared to the system suppliers suggestion?

2. Literature reviewA large body of literature exists on spare parts inventory theory. Two major reviewpapers are Guide et al.s (1997) paper on repairable inventory theory and Kennedyet al.s (2002) paper regarding general spare parts inventories. According to Guide andSrivastava the classical repairable inventory problem and the multi-echelon modelsdates from the military. Guide and Srivastava reviewed 42 papers regardingmulti-echelon models, covering the timeframe from 1968 to 1996. Kennedy et al. have areference list of 61 papers included in their review. Tysseland and Halskau (2007) havea literature review with a special focus on initial provisioning and obsolescencemanagement. This paper pay respect to earlier review papers and focuses especially on

Spare partsoptimization

process

9


4/21

the timeframe from 2001 to 2007. A new focus in this paper is the effort done in findingempirical studies connected to the use and implementation of spare parts inventories.The authors found only five papers on the topic even if papers from before 2001 wereincluded. The last of these five papers, a paper by Zomerdijk and de Vries (2003)extend inventory control into organizational management. Zomerdijk and de Vries(2003) claim that the vast majority of research within inventories originates from thefields of operation research, and thus the concepts and techniques are mainly based onmathematical assumptions. They further claim that despite the value of the availableconcepts, this common background in operation research has its limitations especiallyin implementation, management and practical control of spare parts inventories. In this

review, no further papers were found connecting the specific of spare parts inventorytheory into organizational issues. One literature review paper by Wong et al. (2004)regarding general supply chain coordination problems proved however to beimportant for this research, and this will be further elaborated in the next chapter.

3. Theoretical frameworkThe research questions dictate that the theoretical framework must be both spare partstheory and organizational theory. The spare parts theory should be firmly anchoredwithin organizational theory, because the utilization of the spare parts theory happenswithin an organizational setting. Coordination of the spare parts optimization processis needed both within the organization itself (e.g. between project leader (PL) and theprojects ILS manager called Project Coordinator ILS (PC ILS)) and with the projectssupplier(s). According to van de Ven et al. (referred in Wong et al., 2004), coordination

is defined as the integration or linking together of different parts of an organization toaccomplish a collective set of tasks. Lack of coordination causes many problems, butprimarily affects organizational or economic effectiveness and efficiency (Wong et al.,2004). In the case of this research it is believed that lack of coordination in the spareparts optimization process can lead to at least two main problems; first of all one canget less system effectiveness in terms of low-system availability. Secondly, one can getlow-system efficiency by spending more money then needed on spare parts in order toachieve a preset availability goal. Wong et al. (2004) have looked at the coordinationproblem from three different theoretical standpoints, namely organizational theory,economic organizational theory and operation management theory. All three theoriesrecognize the source of coordination problems as lack of information symmetry, lack ofcentralized decision making, uncertainty and interdependency. Further, especiallywithin economic organizational theory, limited rationality and behavioural issues areidentified as important sources for coordination problems. Based on this, six causes for

coordination problems are included in the research model. It is assumed that whensystems with long life cycles are procured, system owners are faced with the challengeof achieving high-system effectiveness (high performance and high availability) withlow life cycle cost (LCC). If the performance measurement is acceptable, the challengebecomes how to achieve high availability in terms of having as few systems as possiblethat are Not Operational Ready (NOR). NOR as a measure of effectiveness in spareparts optimization is not the best solution. Thus, in literature (Alfredsson, 1997) it hasbeen shown by a chain of equivalence that minimizing NOR is equivalent tominimizing the Number of Back Orders (NBO). One could of course minimize NBO bybuying an infinite amount of spare parts S at cost C. In most procurement projects

IJPDLM39,1

10


5/21

however there will be a budget restriction b. All systems consists of several subsystems, here denoted by k such that a system is a sum ofk 1, . . . K. In the simplestversion with one indenture level and one site the problem can be defined as:

minXK

k1

NBOkSk s:t:XK

k1

CkSk # b Sk $ 0 and integer

Looking at the minimization problem we see that it is not intuitively for anybody to

decide which spare parts in what numbers are best to buy in order to minimize thenumber of back orders given the budget available. The problem is called an integerknapsack problem, and this problem has been solved in literature and spare partsoptimization software such as OPUS10, by utilizing an algorithm called marginalallocation. Further details of the marginal allocation solution and how the expectedNBOs are calculated in OPUS10 will not be given here[3]. In OPUS10 the complexity ofseveral multi-indenture systems being supported by a multi-echelon support system istaken into consideration. In many real life approaches to spare parts selection, someacknowledgement of the knapsack problem is maybe present, but the solution need notbe given by a system-based optimization process. In a publication by Systecon (2002) asimple test case[4] is used to illustrate the challenge of deciding the number of spareparts to buy based on four different approaches. The four approaches are; randomchoice, same of each, constant confidence against stock out and the system approachbased on OPUS10. In the random choice approach a random number of spares from 0 to

10 is chosen for each line replaceable unit (LRU), giving 11[5] possible combinations(hopefully not much used). The second approach is known to be used in procurementprojects in the Norwegian Defence. This approach basically means that you buy thesame of each LRU (often one). The third method is more sophisticated because it takesinto account the utilization factor, failure rate per LRU, number of LRUs per systemand turn-around times (TAT) for repair of the LRU when the expected demand iscalculated. All LRUs are then calculated to the quantity that meets a given confidencelevel (e.g. 95 per cent) for not having a stock out during a certain period of time. Themain challenge with this method is that unit price of the LRU is not included inthe calculation. If unit price for each LRU is the same then this method is close tooptimal. We know however that in real life this is not the case, and the effect of unitprice should be considered, especially since in most cases there is a limited budget forspare parts. Figure 1 shows the results on availability and cost for different stockingalternatives for the three last methods (Systecon, 2002).

As can be seen, the system-based optimization method based in OPUS10 (indicatedas Initial in Figure 1) is by far the best in this hypothetical case. Based on this theresearch questions stated in Chapter 1 becomes very important.

4. Research model and hypothesesThe research model (Figure 2) is based upon the six causes for coordination problems(or coordination factors) extracted from organization theory, economic organizationtheory and operation management theory (Wong et al., 2004).

Addressing the RQ1, it is believed that by studying the different projectscoordination factors one can better understand how the spare parts optimization


process

11


6/21

process has been conducted and whether coordination issues have affected the processand results.

Project uncertainty and interdependency are defined within the physical context ofthe project in question. Project uncertainty can lead to coordination problems in thespare parts optimization process, resulting in less use and diminished results. Lowuncertainty can lead to less coordination problems, which means better spare partsoptimization. Project uncertainty could be studied by looking at the number of tasks inthe project, and the uncertainty connected to the tasks. In this study, spare partsoptimization is looked upon as one large task with associated sub tasks anduncertainty connected to it. In order to conduct a valid spare parts optimization,trustworthy input data in terms of for example validated failure rates are needed.Based on this, uncertainty is firstly defined as whether the system procured can be

acquired so-called commercially off the shelf (COTS), or whether it has to bedeveloped specifically (high-asset specificity). It is assumed that COTS systemsindicate a better possibility for access to needed input data than asset specific systems.This is because COTS systems are most likely to be in use by others and validatedoperational data can more easily be acquired by the project. This will reduce the risk ofdata input error in the optimization. If the system is not in use by others, or failure ratescannot be obtained, the risk of making incorrect calculations will increase. Further thesize of the task including sub tasks will affect uncertainty. A small project with fewLRUs connected to it will be less uncertain than a large and complex project. Regardingproject uncertainty the following hypothesis is proposed:

Figure 1.

Three alternative methodsor spare partsdimensioning

Optimization vs Constant Confid. and Same of Each Availability

0 50,000 1,00,000

Total Investment

0.2

0.4

0.6

0.8

1.0

Same of Each

Const.Config.

Initial

Figure 2.The research model

Spare parts

optimizationResultInformation symmetry

Centralization

UncertaintyInterdependency

Behaviour / Attitude

Rationality

IJPDLM39,1

12


7/21

H1. High-project uncertainty can lead to lower use of spare parts optimizationbased on OPUS10.

Coordination problems based on interdependence are due to scarcity of resources orscarcity of information or a combination of the two (Wong et al., 2004). This dimensionis placed within the frame of physical context because in terms of spare partsoptimization the project in question can either get access to all needed information(e.g. failure rates) through open sources or they must relay on the system supplier. Inthe same direction the project can have access to needed resources, for example interms of personnel with knowledge of spare parts optimization, within their own

organization or they must get this outside their own physical context:H2. Low control over resources and information connected to spare parts

optimization will negatively affect the use of OPUS10.

Governance and organizational structure is closely connected to the research questionon how the optimization based on the system approach has been conducted. Fourpossible reasons for coordination problems that can hamper the optimization processare included within this frame. The four dimensions in question are informationsymmetry, centralization, behavioural issues (attitude) and rationality. Symmetry ofinformation between the PL and the PC ILS is of interest. According to Eisenhardt(1989) information symmetry refers to the principal and agent possessing the sameinformation and information symmetry can thus curb adverse selection. The argumentis that since information systems can inform the principal about what the agent isactually doing, the agent will realize that he cannot deceive the principal. In this case

information symmetry can also be between the project and the main spare partssupplier. According to Wong et al. (2004) the more complete, timely, observable andverifiable information one has, the less possibility for coordination problems and thebetter result. Based on this the following two hypotheses are suggested:

H3. Low-information symmetry between PL and PC ILS regarding spare partsoptimization will negatively affect the optimization.

H4. Low-information symmetry between spare parts provider and projectregarding information needed for spare parts optimization will negativelyaffect the optimization.

According to theory centralized decision making will reduce coordination problems.In this context, centralized decision making could be done through the utilization of acommon concept for spare parts optimization implemented through PRINSIX, and

followed up by key performance indicators (KPI). A decentralized version would be tolet each PL decide how to perform the spare parts optimization process and setting ofspare parts inventory control variables (e.g reorder points):

H5. Decentralization will negatively affect spare parts optimization based onOPUS10.

In this research the concept of rationality is defined as a ratio between the cognitivecapability of the decision maker and the complexity of the problem in question (Heiner,1983 in Wong et al., 2004). The decision maker has absolute rationality when his/hercognitive capabilities completely match the complexity of the problem. Complete


process

13


8/21

rationality rarely happens in an organization; instead rationality is limited or boundeddue to the complexity of the problem (Wong et al., 2004). Bounded rationality may leadto the use of rules of thumb in the decision-making process, such as in the decisionprocess of spare parts:

H6. The more complex the system becomes compared to the decision makercognitive capability, the more likely are the decision maker to use rule ofthumb sparing.

Traditionally project success is measured on the triple constraints; time, budget andoverall quality (Gemunden et al., 2005). However, the triple constraint has often onlyincluded the pure project phases, not the operation and support phase. If for examplethe PL and/or PC ILS think that their future (for example promotion) is based on thefact that the procurement project is finished on time and within the original investmentbudget, their attitude towards the use of spare parts optimization might be lesspositive. This is because the focus in spare parts optimization is on operation andsupport cost and availability in addition to the traditional triple constraint. PLs and PCILS acting in what they think are their self-interest at the expense of the NorwegianDefence overall interest is called adverse selection (Eisenhardt, 1989). However, it ishard to measure such behaviour by questions. Based on this, a substitute in terms ofthe PC ILS general attitude towards the use of LCC-based methods (such as spare partsoptimization based on OPUS10) is used as an indicator. Negative attitude is notnecessarily a sign of bad will, but can be the effect of for example lack of information.The following hypothesis is proposed:

H7. Positive attitude towards the use of LCC and spare parts optimization basedon OPUS10 will positively affect the use.

5. MethodologyThe RQ2 regarding the result of OPUS10-based spare parts optimization leads us inthe direction of a quantitative data analysis and hence the need for quantitative data.The RQ1 (the how question) on the other hand leads us in the direction of case studiesas the preferred research strategy (Yin, 2003). The qualitative data were collectedthrough semi structured individual interviews including sets of structured questionsand by obtaining archival data. An interview guide[6] was developed based on theresearch model, containing 76 questions, where 36 questions were in a structuredformat. Each interview lasted approximately 90 minutes. The intention was tointerview the main responsible for the spare parts optimization process in each of the

projects were OPUS10 had been used since year 2000. As can be seen in the nextchapter, the total number of projects was no more than nine. Some of the responsiblepersons in these nine projects had left the Norwegian Defence, but it was still possibleto get interviews with persons connected to the optimization process in eight out ofthe nine projects. Each interview was taped and main phrases transcribed to paper. Sixinterviews were done face to face and the last three interviews were done by speakerphone. Nine interviews in eight projects were conducted, because in one project tworesponsible persons, one from an earlier stage along with the one responsible todaywere interviewed. The interviews together with archival data were then used toanswer/explain the predefined hypotheses. The quantitative spare parts data were

IJPDLM39,1

14


9/21

collected by getting access to project databases and files along with the managementsystems of the Navy, Army and Norwegian Defence Logistics Organisation (NDLO).

6. The cases and standard spare parts procedureSince OPUS10 already in 1995 had been recommended as the main tool for spare partsoptimization in the Norwegian Defence (PS2000, 1995) one should believe that therewere many projects to get data from. In the Norwegian Defence Project Database thereare currently several hundred projects, with an annual investment amountof approximately ten billion Norwegian Kroner (NoK). In the search for cases,administrators for this database along with those responsible for the current contractwith Systecon AB (OPUS10 provider) were contacted. This lead to contacts with thosevery few (compared the total number of projects running) projects that in the lateryears have used OPUS10 in the spare parts optimization process. More specifically itwas found that no more than approximately nine projects had used OPUS10 since thelate nineties. Since much of the information provided by the Norwegian Defence in theireyes are sensitive, all projects and informants are kept anonymous and givenpseudonyms. Out of the nine projects identified to have used OPUS10 over the last tenyears, data have been collected from eight of them.

As shown in Table I, only two projects had operational data available. Project NAwas part of a large Navy procurement program that started in the 1980s. The projectdeveloped and procured a weapon system in the late 1990s to be used on a class ofNavy vessels. NA was a system specially developed in cooperation between theNorwegian Navy and the Norwegian defence industry. The project had a total budgetof more than 200 million NoK, and approximately ten million (hence in accordance withthe 5 per cent estimate) were targeted for spare parts. Project AA delivered a mainweapon system which in 2004 was considered for a spare parts update since more ofthe systems were to be used in international operations. OPUS10 optimization had alsoearlier been conducted on this system, but the focus of this analysis has been on thenumber of LRUs considered for increase in stock based on the 2004 analysis. The othercase studies were not in the operational phase of their life cycle.

The standard procedure in the Norwegian Defence regarding spare partsdimensioning is that the main system supplier through contractual demands isobliged to provide NDLO with a suggested spare parts list. This spare parts listprovided by the system supplier can in theory be based on any of the methods shortlypresented in Chapter 3. Through a series of spare parts provisioning conferences the

Project pseudonym High cost project(.500 million NoK) Customer Life cycle phase Operationaldata available

NA No Navy Operational YesNB Yes Navy Procurement NoNC Yes Navy Procurement NoAA No Army Operational YesAB No Army Procurement/Operational No/littleAC No Army Procurement/Operational No/littleFA Yes Air force Procurement No

JA No Air services Procurement No

Table I.Case projects

characterization


process

15


10/21

system supplier and NDLO will normally agree upon a final spare parts delivery. It isin this process that the use of a multi-echelon, multi-item, multi-indenture spare partsoptimization method is thought to be of high importance in order to achieve the bestpossible operational availability for the system at the lowest possible cost, or to achieveas much availability as possible for a given budget. According to the informants, it isthe organization within NDLO with technical responsibility for the system in question(NDLO Systems Management Division NDSMD) that is supposed to lead the sparingprocess. The informants in this research confirmed that to the best of their knowledgethe multi-echelon, multi-item, multi-indenture spare parts optimization method wasonly used through the system OPUS10 in Norwegian Defence procurement. They

further confirmed that in the projects that have not used OPUS10, NDSMDrepresentatives would work through the suggested parts list and maybe question someof the suggestions, but basically just draw a line when the budget no longer allowedNDLO to buy more. Based on this one can say that the process in the vast majority ofprocurement projects thus far at best can be called an engineering choice process. Theresult will be affected by how the system supplier has constructed the spare partsproposal. If the supplier has taken into account the element described in Chapter 3under the system-based method or used the constant confidence against stock-outmethod, the system availability will be better that if the supplier has used the same ofeach, engineering choice or random choice method. Empirical data show that a lotof spare parts bought by the NDLO have no movement at all throughout the life time ofthe system they are bought to support. Hence, it is likely to believe that systemsuppliers spare parts suggestions and NDLO buying decisions not necessarily takesinto account system-based criterions such as for example mean time between failures(failure rates). In the projects that use OPUS10 however, the system supplierssuggestion will be evaluated as explained in Chapter 3. After the final list of spares aredecided, the project together with NDLO Supplies Management Division (NDSuMD)are supposed to make sure that all spare parts get a NATO Stock Number (foridentification purposes) and associated inventory control measures (e.g. reorder points

ROPs).

7. Qualitative data analysisTable II is a summary of the findings from the analysis of the qualitative datacollected. In the following, analyses against the predefined hypotheses for each of thecoordination factors are presented. The use of OPUS10 in the nine case studies isdivided into three categories. The first category is coded; complete this means thatthe project has used OPUS10 to optimize the entire system when it comes to spare

parts. Five out of nine project cases falls into this category, including both systemswith operational data (system NA and AA). The second category is coded; partly thismeans that only certain sub-systems within the total system have been optimized withOPUS10. Two systems NB and FA fall into this category. System NB for exampleconsists of two major sub-systems and one of these has been optimized with OPUS10.The last category is; partly to none this means that OPUS10 has been used at onestage in the project development but later dropped. This is the case of system NC.

The findings in the nine case studies to a certain degree support H1 regardinguncertainty. Two of the three case studies that had not used OPUS10 completely hadhigh uncertainty both in terms of being specially designed systems (non-COTS), and

IJPDLM39,1

16


11/21

Uncertainty

Interdependency

Infosymmetry

Centralization

Rationality

Attitude

ProjectCOTS

Task

Info

Resources

InternalE

xternal(Instructions./KPI)(Complexity)P

L

PCILS

Use

NA

High

Low

Low

Medium/low

High

H

igh

Medium

Low

H

igh

High

Complete

NB

High

High

Medium/high

Medium/high

Low

M

edium

Low

High

L

ow

High

Partly

NC

High

High

Medium/high

Medium/high

Low

M

edium

Low

High

L

ow

Medium/

high

Partlyto

more

AA

Medium

Low

Low

Medium/low

High

H

igh

Low

Low

H

igh

High

Complete

AB

Medium

Low

Low

Medium/low

MediumH

igh

Low

Low

M

edium

High

Complete

AC

Low

Low

Low

Medium

MediumH

igh

Low

Low

M

edium

High

Complete

FA

Medium

Low/medium

Medium

Medium

MediumM

edium/

high

Low

Medium

M

edium

High

Partly

JA

Medium

Medium

Low

Medium/low

High

H

igh

Low

Low

H

igh

High

Complete

Ideal

Low

Low

Low

Low

High

H

igh

High

Low

H

igh

High

Table II.Summary of the

qualitative data analysis


process

17


12/21

having a very high-task complexity in terms of being large specially designed systemswith a vast amount of sub-systems and LRUs attached to the systems. Both system NBand NC are systems for the Navy with very large investment budgets (multi billionprojects) and a very high complexity. The last project that has not used OPUS10completely is also a very complex system, but it has a somewhat lower uncertainty interms of the COTS dimension because several nations is procuring this systemtogether and because the type of industry (aircraft industry) delivering the system hasa long history of working with for example failure rates needed to do the spare partsoptimization. The projects with complete use of OPUS10, have all except one (JA) beenidentified to have low uncertainty in terms of the task dimension. This is because thecomplexity is terms of number of systems, LRUs, support organization and so on is notvery large. The task complexity for system JA was identified to be somewhat higherthan the others but still only low to medium. Compared to the hypothesis the finding ofhigh uncertainty in terms of a very asset specific system in the case of project NA wassomewhat surprising. However, this high uncertainty is in this case offset by alow-task uncertainty and almost ideal findings in four of the five other dimensionsreducing coordination problems.

In connection with H2, it was found that scarcity of resources in terms of personnelwith knowledge of spare parts optimization were evident in all projects. It is howeverimportant to notice that in the studied projects at least some resources are availablewhile in the large amount of projects found to not have used OPUS10 one can predictthat recourses are almost non present. Seven out the eight cases reported that no morethan one person in the project had good knowledge of spare parts optimization based

on the system approach. The NC project had at some times more than one person withthis knowledge but this project is a very large project (one of the largest in the NDLOportfolio) with high complexity and a very high need for skilled ILS project members.All projects reported that they used resources from outside the project itself to run theOPUS10 software. Before 2002 defence procurements were conducted by the differentservices (Army, Navy and Air Force) in their own material commands. From 2002 thethree commands were organizationally joined in NDLO. However, the three physicalsites[7] of the old commands were kept, and hence most Army projects are run from theold Army site, Navy projects from the Navy site and Air Force projects from the AirForce site. Before 2002 all three material commands had their own ILS communitywhere spare parts optimization, at least theoretically, was a part of the portfolio. At thetime of consolidation the organizational elements in Bergen (Navy) and Kjeller(Air Force) were terminated and a common community was established based in the oldArmy site at Kolsas. One project of the eight studied was finished before the

consolidation of the ILS communities (project NA). In this project, resources from theNavy ILS community together with resources from Systecon AB (the OPUS10provider) ran the optimization. They also highly involved the organizations withtechnical and supply responsibilities towards the system in the optimization process aswell as the system supplier. The informant in the NA project said that this was veryimportant because such an iterative process removed misunderstanding an errors thatotherwise would not have been detected. After 2002 the ILS community situated at theArmy site have had limited amount of resources. Through the study it was found thatthree out of four projects with complete use of OPUS10 are ran out of the old Army siteand thus have had physically closer access to the ILS community. The informant in the

IJPDLM39,1

18


13/21

Air Force project (project FA) reported that he had not heard of the new consolidatedILS community before 2006, and hence thought that such resources were not availableafter the termination of the Air Force command ILS community in 2002. However, bothproject NC and JA which are both located in the old Navy site had used resources fromthe common ILS community and especially project JA reported that this was what theyneeded in their process. It is interesting to see that all of the projects that have done acomplete optimization reported that they had low scarcity of information, which is inaccordance with the hypothesis. Project NB and NC reported that scarcity ofinformation, especially for parts of the system made it impossible to conduct acomplete optimization.

The informants in NB, NC and FA all reported that they experienced lowinformation symmetry between the PL and PC ILS regarding spare parts optimization.All three reported that the leadership within the project had low knowledge and not tomuch interest in the spare parts optimization process and result, at least not early inthe project. On the opposite spectrum the informant in project NA reported that the PLwas very interested in the process and result because the PL had committed todelivering certain system availability in operations. Basically the findings from thecases studies are in line with H3. For H4 the findings from the case studies are verywell aligned with the proposition. In those case study projects that have not usedOPUS10 completely the informants reported that a major reason for this is the problemof getting reliable data from the system supplier regarding certain areas orsub-systems of the total delivery. For example, in project NB, the informant stated thatin the propulsion part of the project the supplier were not able to provide failure rates

for the propulsion system. The informant in this project and in the NC project reportedthat some times the requirements against the system provider were not well enoughstated in the contract. Hence, if the system provider did not want to come up with theneeded data there were no contractual way of making him do so. To sum up, findingfrom the case studies do support the two hypotheses put forward regardinginformation symmetry.

There is only one project of the eight cases studied that even come close to thetheoretical concept of a centralized methodology and decisions reflected in H5. Theproject in question is the Navy project NA, which was completed, based in the oldorganization with a separate Navy ILS community deciding, or at least trying to decidehow the Navy projects should conduct spare parts optimization. Talking to theinformants from the other projects it is clear that a centralized decided process for howthe spare parts optimization shall be done is not present. A proposed concept writtenby the ILS community situated at Kolsas was found[8], but this concept is not

approved by NDLO and hence not necessary for the different projects to follow. Onecould claim that support for the hypothesis is actually found because so fewprocurement projects actually use OPUS10.

In H6 it was stated that the more complex the system becomes compared to thedecision maker cognitive capability, the more likely are the decision maker to use ruleof thumb sparing. Rule of thumb decision making in this context is that the projectsaccepts the systems providers proposal either totally or by using engineering judgmenton each of the spares suggested, and making corrections based upon this. In three ofthe eight cases studied the total complexity of the system delivery was very large(system NB, NC and FA). Nevertheless, the rationality is coded as high in NB and NC


process

19


14/21

but medium in FA. The reason for FA being coded medium has not to do with thecomplexity being lower but that finding that the cognitive capability of those involvedin the project was found to be higher than in the other two projects. The FA project is aproject where several nations work together on the procurement. In this way NDLOhas been able to utilize specialists in spare parts optimization connected to one of theother nations project team. In this way the ratio between the complexities of theproblem compared to the cognitive capability of the project has been lowered comparedto the large and complex NB and NC projects that are purely Norwegian. In the rest ofthe projects the ratio between complexity and cognitive capability is lower. Findingtowards the concept of rationality seems to fit the hypothesis put forward.

The findings connected toH7are divided into PLs attitude and PC ILS attitude. Forthe PC ILS informants the findings were that all were positive to spare partsoptimization based on OPUS10. One of the informants in the NC project said that hehad had some bad experience with OPUS10 optimization due to lack of reliable inputdata, and based on this he was somewhat reluctant to use the system. All PC ILSinformants were also asked structured questions regarding their attitude towards theuse of LCC-based procurement decisions (OPUS10-based spare parts optimization canbe part of such an LCC analysis) identical to those given to all PLs in the NorwegianDefence in a research study from 2007 (Tysseland, 2007). Since items and summedscale regarding attitude was found to be both reliable and valid in the Tysselands(2007) paper the exact same items and summed scale were used in order to test whetherthere was a statistical significant difference in the attitude towards LCC-basedprocurement decisions between PLs and the PC ILSs included in this study. The meanfor the PLs (sample size 78) was 3.24[5], with a standard deviation of 0.95. The mean forthe PC ILSs (sample size nine) was 4.56, with a standard deviation of 0.64. By usingSPSS[9], both the parametric t-test and the nonparametric Mann-Whitney test showedthat there was a statistically significant difference in attitude between the two groups.This is clearly in line with the qualitative findings were the PC ILS informants reportson a general basis that their PLs are not that interested in LCC-based decisions andspare parts optimization based on the system approach. However, the data also showthat in those cases that the PL has what can be termed as a very good attitude such asin case study project, NA, AA and JA it is more likely that the project has used OPUS10on all parts of the spare parts process. In the cases studied, the PC ILSs attitude has nothad an large impact on whether OPUS10 is used completely or not. This is notsurprising, since they have chosen to use OPUS10 in the first place. If all projects notusing OPUS10 had been included in the study this finding might have been different.However, the attitude of the PL seems to affect the use of OPUS10 even in the projects

were OPUS10 is used, and thus the hypothesis is partly supported.

8. Quantitative data analysisIn RQ2, the question was if empirical data could show that the multi-echelon,multi-item, multi-indenture method based in OPUS10 is better than other methods(e.g. engineering choice and same of each) in terms of system availability and spareparts investment cost. It is important to notice that projects not using OPUS10 will,according to the informants, either accept the system suppliers suggestion as it is ormake a more or less informed change. The suppliers suggestion could theoretically bebased on a multi-echelon, multi-item, multi-indenture method, but according to the

IJPDLM39,1

20


15/21

informants this normally is not the case. The informant reported that the supplierssuggestions most likely were based on a same of each approach or an engineeringchoice approach. Again according to the informants, the main motivation for thesystem supplier is to sell as many spare parts as possible. In order to get as close tothe answer of the research question as possible with the available data, it was decidedto look at the difference between system supplier suggestion and the case projects finaldecision based on OPUS10. As stated before only two out of eight studied projects werein the operation phase of the life cycle, but several of the eight projects had someevaluation of cost and availability.

In Figure 3 the supplier suggested spare parts investment cost versus the

OPUS10-based decision is shown for four out of eight case projects. The reason for notincluding project NA is that initial spare parts investment savings was not a part of theNA project. Project NA had received a spare parts proposal of approximately tenmillion NoK. The suppliers suggestion was in this case a same of each (actually oneof each) solution. In Chapter 3 it has been shown that the same (one) of each solutionwill give lower system availability at the same cost compared to the system approachbased on OPUS10. Project NA chose to use approximately ten million NoK on sparesand focused on increasing availability compared to what the system suppliers initialsuggestion gave. Further case project AC is not included because in project AC, NDLOworked together with the supplier on the spare parts optimization and as such neverreceived a separate spare parts suggestion form the supplier. Finally, in the two largeprojects NC and FA it has not been possible to obtain detailed enough information toquantify the savings.

In project NB only the weapon part of the total system was optimized based onOPUS10. For the weapon system part of the project the initial suppliers suggestion forspare parts amounted to 88.4 million NoK. When this combination of spares was fedinto OPUS10, the system returned an availability of 78 per cent. Project NB then usedOPUS10 to run an optimization without forcing a given combination of spares, butsubject to a budget of 88.4 million NoK. This run gave a combination of spares whichwould according to OPUS10 increase availability to 96 per cent. Project NB knewhowever that they did not have 88.4 million NoK to spend on an initial spare partspackage. First they therefore tried to see what an engineering choice reduction(reducing the list based on experience alone by experts from NDSMD) of the OPUS10optimal package with a budget of 88.4 million NoK would mean in terms of systemavailability. When reducing the package the engineers did not calculate the cost of thereduced list. The engineering choice list was then fed into OPUS10, which showed that

Figure 3.Supplier suggested spare

parts investment costversus OPUS10-based

decision

Spare parts cost evaluation

0

20,000,000

40,000 000

60,000,000

80,000,000

100,000,000

NB AA AB JA

System

CostinNoK

Supplier suggestion

Opus based decision


process

21


16/21

the result would be a spare parts investment cost of 61 million NoK with an associatedsystem availability of 55.4 per cent. The total budget available was however onlyapproximately 36 million NoK. Project NB rejected the NDSMD engineering choicesuggestion and let OPUS10 run an optimization subject to the budget limit of 36million. This final effort gave an availability of 62 per cent. To sum up NDLO were ableto reduce the cost of spares with 58.7 per cent compared to the system suppliers initialsuggestion still only loosing 16 per cent points in availability. In project AA thesuppliers suggestion for increase in spares amounted to 11.5 million, which after thespare parts optimization with OPUS10 was reduced to 5.1 million NoK (saving 55.7per cent). In project AB the suppliers suggestion was approximately 50 million forspare parts. The informant in the project said that after running OPUS10 they wereable to get the desired availability with a spare parts cost of five million. However, byincreasing the failure rates given from the supplier with the concept of operatornegligence, damage and misuse, the project ended up with spare parts buy of tenmillion. A saving of 80 per cent compared to the suppliers suggestion. Project JA is asmaller project in terms of spare parts need, still the saving compared to the supplierssuggestion (thought to be based on an engineering choice solution) was reported to be1.48 million or 46.8 per cent.

The findings clearly indicates that suppliers do not necessarily take into account forexample the mean time between failure (MTBF) on sub systems and TAT on reparableparts when they suggest initial spare parts lists. The incentive seen from the systemsuppliers side is of course, to sell as much spare parts as possible. The problem seenfrom the Norwegian Defences side is that spare parts are bought that might not be

used at all in the life time of the system it is bought for (due to MTBF being far higherthat the life time of the system it is used within) or that they are bought in excessquantities. As mentioned before, in a report from 1995 (PS2000) it was estimated thatthe Norwegian Defence at that time bought spare parts for 120 million NoK every yearthat would most likely not be used in the life time of the system they were bought tosupport. This analysis clearly indicates that this is still a problem. The cost savings inthe four projects that was possible to get detailed information from has, at leasttheoretically, been on the average 60.3 per cent. In three out of four projects theavailability percentage has been kept unchanged or improved even if the spare partsinvestment cost were this much reduced. In the last project (NB) the reduction inavailability was minimal compared to the reduction in spare parts investment cost.Based on the informants claims that the suppliers suggestions are most likely eitherput together randomly, by engineering choice or as a same of each solution theempirical findings in the cases where the multi-echelon, multi-item, multi-indenture

method based on OPUS10 have been used clearly indicates that this method isbetter than other approaches in terms of spare parts investment cost and systemavailability.

The question now becomes whether real life operational data can sustain the claimsand calculations done in the projects. As mentioned before only two of the case projectshad operational data, namely NA and AA. Data from these two projects have beenevaluated over the period from the start of 2004 through the end of 2007. Average fillrate for all Navy weapon system from 2004 to 2007 is 88.2 per cent. The development infill rate for the total group has been steadily declining from 93 per cent in 2004 to82 per cent in 2007.

IJPDLM39,1

22


17/21

Figure 4 first shows a sample of Navy systems average fill rates in the period,including system NA. Secondly, the figure shows the development in fill rate forsystem NA from 2004 to 2007. The total inventory for system NA today consists of 184SKUs. The fill rate from central inventory has so far been very good with an average of95.75 per cent. Somewhat surprising is the fact that out of the 35 SKUs that have haddemand, 13 (37 per cent) has no longer stock. None of the 13 has so far had rest order,but of course, with S 0 for all 13, the chance for rest order and reduced systemavailability is definitely very present. It is NDSuMD who is responsible for inventorymanagement and the NDSMD informants with technical responsibility for the NAsystem were very surprised by the finding when presented with it. According tostandard procedure NDSuMD gets paid each time a customer (e.g. Navy) use a sparepart. NDSuMD is supposed to use this payment to reset the inventory to the amountdecided by NDSMD.

In Figure 5 the average fill rate for several Army systems including system AAfrom 2004 to 2007 is given in addition to the year by year fill rate for system AA. Theavailability requirement in project NA was 95 per cent.

According to NDLOs Army inventory system, the number of SKUs connected tosystem AA is 1040. Out of these 1,040 SKUs, 665 had stock at 3 March 2008. However,only 479 SKUs had order points in the management system (a NDSuMDresponsibility). Out of these 90 had both the setting for maximum inventory positionand order point (defined as minimum inventory position) set to zero. The questionbecomes whether the zero setting for these 90 SKUs are real settings or just a

Figure 4.Navy systems average fill

rate (2004-2007) andsystem NA fill rate per

year

Navy systems average fill rate in percent 04-07

90 91

74 74

89

96

65

70

75

80

85

90

95

100

Navy

EquipmentA

Navy

EquipmentB

Weapon

systemNX

Weapon

systemNY

Weapon

systemNZ

Weapon

systemNA

Fill rate in percent System NA

100 100

83

100

40

50

60

70

80

90

100

2004 2005 2006 2007

Figure 5.Army systems average fill

rate (2004-2007) andsystem AA fill rate per

year

Army systems average fill rate in percent 04-07

89

92

82

88 89 90

65

70

75

80

85

90

95

100

Army

VehicleA

Army

VehicleB

Weapon

SystemAY

Weapon

SystemAY

Weapon

SystemAA

Weapon

SystemAX

Fill rate in percent System AA

95100 100

55

40

50

60

70

80

90

100

2004 2005 2006 2007


process

23


18/21

system default. One can speculate that actually only 479-90 389 SKUs have realinventory settings. The analysis showed that 697 SKUs have had orders in the period2004-2007, and out of these 435 SKUs have had rest orders at least at one point. For theupdate job in 2004, 273 SKUs were considered for an update of settings (a change in thenumber of spare parts based on the new situation). Out of the suppliers suggestion of273 items to be increased, the AA project based on the OPUS10 analysis removed 101items. By studying the development in fill rate an interesting development occurred.As can be seen in the Figure 5 the fill rate for the first year after the optimization is verygood with 95 per cent, and than even better for year two and three with 100 per cent fillrate for both years. In 2007 a dramatic change in fill rate is observed. The data analysisrevealed the simple but still puzzling reason for this dramatic change in fill rate. Forthe SKUs included in the OPUS10-based optimization in 2004, NDLO have after theprocurement of additional spares not sat order point or other decision variables(e.g. maximum allowed stock) in the spare parts management system that System AAis managed by. The result became that no warning was given before the differentSKUs reached a stock of zero with the given result of back orders.

To sum up the quantitative data analysis, the empirical findings clearly indicatethat the multi-echelon, multi-item, multi-indenture method based on OPUS10 is betterthan other methods in terms of both system availability and spare parts investmentcost. However, the data analysis also showed that poor follow on spare partsmanagement can in the longer run ruin system availability.

9. Conclusion, limitation and future researchThe empirical findings clearly support the theoretical claims of the multi-echelon,multi-item, multi-indenture method based on OPUS10 being better than otherapproaches (same of each and engineering choice) in terms of system availability andspare parts investment cost. The two cases with operational data (NA and AA) eitherperformed as good as, or better than, comparable systems. The dip in fill rate of systemAA is an interesting observation in it self because this is due to an organizationalinventory management issue rather than the result of a poor spare parts optimization.In project NA and AA, cost saving was only an issue in project AA with 55.7 per centsaving over the system suppliers suggestion. This saving in spare parts investmenthas not reduced availability of the system compared to systems spared with otherapproaches. If the availability of system AA is representative the large reduction inspare parts compared to the system suppliers suggestion clearly indicates that inthe non OPUS10-based projects where the suppliers suggestion is accepted a

considerable amount of spare parts are bought that actually are not needed. Furtherwhen the suppliers suggestion is reduced with an engineering choice approach by theNorwegian Defence such as in project NB, the chances of buying wrong spares andremoving the ones they need are still very present. The reduction in spare partsinvestment cost based on the use of OPUS10 identified in this study (on average 60.3per cent) is very interesting, especially since empirical operational data show thatsystem availability is not reduced. This finding should warrant managerial action.With a conservative approximation of spare parts cost to 3.5 per cent of the totalNorwegian Defence procurement budget, and a reduction of the saving potential to40 per cent, the annual saving in procurement of not needed spare parts would still be

IJPDLM39,1

24


19/21

140 million NoK. Total LCC of these not needed spare parts would be even higherconsidering the costs of holding unneeded inventory.

Very few projects had actually used OPUS10 since it was decided to be theNorwegian Defence standard system for spare parts optimization back in 1995. It isbelieved that this lack of usage is not due to the method and system it self but to howthe different coordination factors are observed and solved by the projects in question.The findings from the qualitative part of the study gave some interesting findings withmanagerial implications. If the Norwegian Defence really wants the system approachbased on OPUS10 to be used completely, the project in question should be mapped outbased on the research model of this study, in order to get as little coordinationproblems as possible. Three problem areas were identified to be the most challengingin the projects that have used OPUS10. These problem areas are probably even biggerin the vast majority of projects that have not used OPUS10. The three areas in questionare; lack of resources; lack of centralized instructions and finally; the PLs attitude. Allstudied projects reported that they had limited resources in terms of personnel with theneeded ILS competency. Given the potential for savings identified, it is somewhatsurprising that NDLO has not focused more on this area. Further all projects reportedlack of one common concept for spare parts optimization as well as to little focus inPRINSIX on the subject. Along with an increase in resources, it is thus believed that acommon concept should be developed, approved and implemented by the NDLO andsustained by PRINSIX. Further the concept must be followed up by KPIs primarilyaimed at the PLs. It is believed that use of KPIs to a certain degree will address thesomewhat low attitude found from the PL towards system-based optimization.

Another way of addressing the attitude issue is to build knowledge within theprocurement community concerning the topic (Tysseland, 2007).Not being able to statistically accept or reject the hypothesis presented in the data

analysis can be argued to be a limitation to this research. The research could bedeveloped by using a questionnaire to reach as many projects within the procurementcommunity as possible. However, it can be also be argued in the opposite direction thatthe use of quantitative data have presented more indebt data then a questionnairecould have given. Further the cases are restricted within the Norwegian Defence andhence generalization of both qualitative and quantitative findings is at bestquestionable and probably not fruitful. The area of research though should be ofinterest to other organizations and industries and the research model could easily betransferred and utilized in the study of other industries who utilize a structuredapproach to procurement including spare parts optimization. In this way a moregeneral template or process map, based on the extension of spare parts theory into

organizational theory could be developed.

Notes

1. The total defence budget was plus 30 billion NoK in 2007.

2. The ten elements in question are: maintenance planning; supply support (mainly spare partsprovisioning and management); design interface; packaging, handling, storage andtransportation; manpower and personnel; support equipment; technical data; training andtraining support; facilities and computer resources support.

3. The interested reader could, e.g. consult Alfredsson (1997).


process

25


20/21

4. The case consisted of 50 systems, one echelon, ten LRUs (one indenture level, five LRUs withtwo per system) with associated, system utilization, failure rates and turn-around times forLRU repair. One of the LRUs is more than ten times as expensive as the others.

5. On a scale from one to five with end points fits very badly and fits very well.

6. The guide is in Norwegian since all informants were Norwegian. It is possible to get a copyof the interview guide by mailing the author.

7. The Army site is located at Kolsas just outside Oslo, the Navy site is located in Bergen andthe Air Force site is at Kjeller North of Oslo.

8. The concept is in Norwegian and a non approved version can be obtained from the author of

the paper.9. SPSS is short for the Statistical Package for the Social Sciences, see www.spss.com

References

Alfredsson, P. (1997), On the optimization of support systems, doctoral thesis, Department ofMathematics, Royal Institute of Technology, Stockholm.

Blanchard, B.S. (2004), Logistics Engineering and Management, 6th ed., Prentice-Hall, UpperSaddle River, NJ.

de Vries, J. (2005), The complex relationship between inventory control and organisationalsetting: theory and practice, International Journal of Production Economics, Vol. 93-94,pp. 273-84.

Eisenhardt, K.M. (1989), Agency theory: an assessment and review, Academy of Management

Review, Vol. 14 No. 1, pp. 57-74.Farmer, M.E., Fritchman, G.E. and Farkas, K.J. (2003), Supporting the fleet in the 21st

century: evolutionary acquisition and logistics, Air Force Journal of Logistics, Vol. 27No. 1, p. 28.

Gemunden, G., Salomo, S. and Krieger, A. (2005), The influence of project autonomy on projectsuccess, International Journal of Project Management, Vol. 23, pp. 366-73.

Guide, V. Jr, Daniel, R. and Srivastava, R. (1997), Repairable inventory theory: models andapplications, European Journal of Operational Research, Vol. 102, pp. 1-20.

Kennedy, W.J., Patterson, W.J. and Fredendall, L.D. (2002), An overview of recent literature onspare parts inventories,International Journal of Production Economics, Vol. 76, pp. 201-15.

Kumar, U.D., Crocker, J., Knezevic, J. and El-Haram, M. (2000), Reliability, Maintenance andLogistic Support A Life Cycle Approach, Kluwer Academic Publishers, Boston, MA.

PRINSIX (2008), available at: www.prinsix.no (accessed February 2008).

PS2000 (1995), Optimalt frstegangs reservedelsopplegg, Norges Tekniske Hgskole,Universitetet I Trondheim, Trondheim.

Systecon (2002), OPUS10 Getting Started, Systecon AB, Stockholm, received at a OPUS10 BasicCourse April 2005 in Stockholm, Sweden.

Tysseland, B.E. (2007), Life cycle cost based procurement decisions a case study of NorwegianDefence procurement projects,International Journal of Project Management, Vol. 26 No. 4,pp. 55-64.

Tysseland, B.E. and Halskau, H. (2007), Spare parts inventory a literature review with focuson initial provisioning and obsolescence management, The 19th Annual NOFOMAConference Proceedings, pp. 1075-91.

IJPDLM39,1

26


21/21

Wong, C.Y., Hvolby, H-H. and Johansen, J. (2004), Supply chain coordination problems: literaturereview from organizational, economics and operations perspectives, Working PaperNo. 08-04, Center for Industrial Production, Aalborg University, Aalborg, pp. 1-23.

Yin, R.K. (2003), Case Study Research. Design and Methods, 3rd ed., Sage, Thousands Oaks, CA.

Zomerdijk, L.G. and de Vries, J. (2003), An organizational perspective on inventory control:theory and a case study, International Journal of Production Economics, Vol. 81/82,pp. 173-83.

Corresponding authorBernt E. Tysseland can be contacted at: [email protected]


process

27

To purchase reprints of this article please e-mail: [email protected] visit our web site for further details: www.emeraldinsight.com/reprints

Date post:	05-Apr-2018
Category:	Documents
Upload:	minakshi-sarma
View:	230 times
Download:	0 times

Spare Parts[1]

Documents